K2-Horizon-32B has 34.779 billion total parameters. The selected Q4_K_M GGUF file is 21.076 GB.
Data snapshot: . 3 published tests in this catalog.
Model file and details
Provider
IFM
Total parameters
34,779,304,960
Exact file size
21,075,788,992 bytes
Quantization / precision
Q4_K_M
Weight repository
mradermacher/K2-Horizon-32B-GGUF
Original model repository
IFM/K2-Horizon-32B
Weight revision
96158cba4289b873548863b48b29169f7ac4affe
Model revision
4b05c13ad1972e8f470ab93ff90c3e90010ce592
Runtime notes
Confirm model architecture and quantization support in your runner.
Recorded release date
2026-09-01
Date evidence
The Apache-2.0 source repository was created on 1 September 2026. A complete independent Q4_K_M conversion appeared on 14 October and reached revision 96158cba4289b873548863b48b29169f7ac4affe on 15 October. IFM describes the source checkpoint as a dense reasoning model with native 524,288-token context, not as a small-memory model; the selected 21.08 GB file does not include runtime, KV-cache, context or output headroom.
These scores were compiled from the linked sources. These published scores were not independently tested here. Different prompts, precision, thinking budgets and tools can change results.
GPQA Diamond
82.8 / 100
Hard science questions. This is a science reasoning test, not an overall intelligence score.
Benchmark efficiency: 3.929 published score points per GB of selected weights. This ratio is not an inference-speed measure; check the test settings to see whether the score covers the selected quantization.
Test settings: IFM-reported GPQA Diamond score for the BF16 source checkpoint in high-reasoning mode with temperature 1.0 and top_p 0.95.
Publisher-run source-checkpoint evaluation, not an independent test of the selected community Q4_K_M. Reasoning budget, prompts, harness and sampling affect results; do not transfer 82.8 to the quantized artifact as measured quality.
Writing code to solve scientific problems. Published results can depend on the test scaffold.
Benchmark efficiency: 1.798 published score points per GB of selected weights. This ratio is not an inference-speed measure; check the test settings to see whether the score covers the selected quantization.
Test settings: IFM-reported SciCode score for the BF16 source checkpoint in high-reasoning mode with temperature 1.0 and top_p 0.95.
Publisher-run source-checkpoint result, not an independent or quantized-artifact measurement. Scientific-coding scaffolds, tool access, generation budget and evaluation harness can materially change the score.
Agentic terminal-use task success. Publisher-reported percentages are retained, but agent scaffold, tools, timeouts and reasoning budget can materially affect results.
Benchmark efficiency: 2.97 published score points per GB of selected weights. This ratio is not an inference-speed measure; check the test settings to see whether the score covers the selected quantization.
Test settings: IFM-reported Terminal-Bench 2.1 score for the BF16 source checkpoint in high-reasoning mode with temperature 1.0 and top_p 0.95.
Publisher-run source-checkpoint result. Agent scaffold, tools, timeouts, reasoning budget and environment materially affect Terminal-Bench; the selected Q4_K_M was not retested. The conversion card supplies files but no pinned runtime version, speed test or quantization-fidelity result, so local compatibility and throughput remain unverified.
Context and version evidence: The repository and file name identify the selected model. Revision hashes are shown above when recorded. Tested context length, full hardware configuration and run date are not consistently recorded in this snapshot. Only settings explicitly stated in each benchmark record are known.
The 21.076 GB figure covers model weights in this file. It excludes the context cache, runtime, operating system and any separate multimodal files. It is not a measured RAM requirement or a token-generation speed.
These complete configurations pass the initial weight-size check for the selected 21.08 GB file. Allow extra memory for context and software. Check the model’s runtime notes and your inference software before buying.
Listing evidence: 2026-09-25. Amazon.com base price visible in an owner-supplied screenshot reviewed 25 Sep 2026; capture time not supplied. Dated reference, not a live quote. Excludes card signup offers, coupons, tax and shipping.
Listing evidence: 2026-10-02. Apple US launch price for the base M5 Max Mac Studio, published 22 September 2026: USD 2,499. Dated manufacturer reference, not an Amazon offer; tax and shipping may differ.
Listing evidence: 2026-09-24. GMKtec, EVO-X2 / 64GB / 1TB, recorded 22 Sep 2026; exact ASIN capacity identified 24 Sep. This is a dated reference, not a current Amazon price; tax, shipping and bundle contents may differ.
Memory matches are estimates, not tested compatibility or speed. As an Amazon Associate I earn from qualifying purchases.
Confirm model architecture and quantization support in your runner. The selected weights use 21.076 GB; this excludes context cache, runtime buffers and other programs.
These are the first listed tiers that pass our reserve policy, not measured minimum requirements. System RAM and VRAM remain separate. Use the memory tool to change the reserve for your context and runner.